So now that I have a BMCC to play with... I tried shooting the same shot with different ISO settings in raw. I then grabbed a DNG from each ISO region and loaded them into Photoshop. On import I clicked "Auto". I then applied -2 from the exposure that the importer chose to get a sense of clipped highlights and in separate set of imports added +2 to the exposure the importer chose to get an idea of grain.
Given that the shots were identical with the exception of ISO on the camera, Photoshop's DNG importer should(?) have exposed the shots so that the output was identical regardless of ISO (assuming analog gain is static).
But this didn't happen.
I couldn't discern any differences between the 400 and 800 ISO shots but there seemed to be a difference between both the 1600 and 400/800 shots and the 200 and 400/800 shots.
This is an experiment, by all means point out flaws in my logic please
. I don't know the answer to the gain question.
Initial observations (all comparing 200 to 1600) :
The grain in the 200 ISO shots seems to be coarser (and have more grain artifacts).
Highlights seem to clip more at 1600.
In the levels dialog less shadow detail seems to be present in the 200 ISO shot.
In the levels dialog less highlight detail seems to be present in the 200 ISO shot.
That said, the differences are very small.
It would be great, just for clarity, to know whether or not ISO does shift analog gain.
I'll upload the DNGs if anyone is interested.
I should also point out that the manual makes it clear that a difference exists between 400 and 800 ISO :
The optimum ISO setting for the Blackmagic Cinema Camera is 800ASA.
In bright conditions 400ASA would be best to record richer colors.
If there is no difference in output, why would 400ISO produce richer colours/colors ?
Finally! The Zebra on the camera's display changes depending on what ISO is selected... it has been stated previously that Zebra represents the absolute sensor limit... which implies that analog gain is being applied if ISO affects it.